Thomas Robitaille

I agree it would be nice to have this, although based on previous experience with OpenMP in other Python packages, I think I'd want this to be opt-in (rather than...

@bastibe - would you be interested in opening a pull request to add these options? I think that the default values that Numpy use mean that this won't add much...

I decided to return doubles for consistency with Numpy, but we could potentially have an option to choose the return type (at least maybe choosing between double and long?). All...

@manodeep - Thanks! I don't have Windows locally, but you can try and open a PR, and there will be an AppVeyor build in which you can check any error...

I wonder if this changed in Numpy since I wrote this package. I'd certainly be up for returning integer values from the plain histogram and doubles from the weighted one,...

The current benchmarks shown in the README are against a later version than 1.10, so this is still quite a bit faster than numpy. I'm not sure if you are...

The code should normally cast to double automatically here: https://github.com/astrofrog/fast-histogram/blob/master/fast_histogram/_histogram_core.c#L111 but it might be this doesn't work for long double somehow. If anyone wants to help investigate, please feel free...

I haven't had a chance to investigate

Failure with the 2D tests: ``` (size=5, nx=1, xmin=0.0, xmax=84.17833763374462, ny=1, ymin=-999999999.9999989, ymax=0.0, weights=False) ```

Failure with 2D tests (same as above but with dtype): ``` Falsifying example: test_2d_compare_with_numpy(size=5, nx=1, xmin=0.0, xmax=84.17833763374462, ny=1, ymin=-999999999.9999989, ymax=0.0, weights=False, dtype='